The Global 2000 Revenue Engine —Fireside Chat with Ishan Mukherjee & Dina Habib OMara

Julia Nimchinski:
Welcome, Ishan and Dina. We’re zooming into enterprise-scale GTM transformation. Dina Habibomara from Microsoft will be leading this discussion, and she’s joined by Ishan Mukharaji, co-founder and CEO of Rox. What a treat! What’s in your agenda, go ask, Ishan.

Ishan Mukherjee:
Yeah, Dean has an amazing kind of lineup, excited to kind of dive into, how real, kind of, agents are on… in the Global 2000, and we’ve been kind of building this for about a couple of years, and there’s been, like, insane market pulls, so excited to dive in.

Julia Nimchinski:
Awesome. Before we dive in, one question we ask all of the panelists is, what’s… what are your favorite tools these days, Ishan? Obviously, rocks.

Ishan Mukherjee:
Yeah, we don’t have any other software in the company. We run the whole business on rocks, which is… which has been amazing to do kind of dog food and kind of build it up. Outside of that, yeah, definitely Cloud Code, like, I am kind of an engineer, and it’s pretty amazing to see how much Cloud Code can do, so I would have to say Cloud Code.

Julia Nimchinski:
Awesome, Dina, how about yourself?

Dina Habib OMara:
Yes, well, I’m… I have… representing Microsoft, I would say, definitely from a frontier transformation perspective, Copilot, we have also our Frontier, what we call the control plane, where you can see all the other tools that Are being used from agents, but absolutely, definitely, the, Agentic AI is here. It is in our workflows, and we can’t wait to, dive into this with everybody.

Julia Nimchinski:
Awesome, let’s do it.

Dina Habib OMara:
All right, thank you, Julia. Thanks. Hi, Ishan, good to see you again.

Ishan Mukherjee:
Yeah, great to see you, and thanks for doing this.

Dina Habib OMara:
Yeah, absolutely. First, I have to call out for everybody, ROX, your company, was the Gartner Cool Vendor Award recipient. So, just want to congratulate you. I think what we have to talk about is going to be very cool today. Specifically, there is a statistic from Gartner I want to call out. It is a shift. A development, a significant shift in how organizations are approaching AI and automation. I mean, it is now becoming crucial. I just heard from the last panel, and it’s spot on. It is not experimentation anymore. That was 2025. We are now in full operation mode, workflow mode, into our platform mode, integrating AI agents. into, becoming crucial roles and team members helping us to fulfill. And in this case, as we talk about the global 2000 revenue engine. It’s about driving revenue. And connecting a lot. So, I wanted to just kind of lay that foundation out, let everyone know we are going to be talking about Agentic AI in our operating systems, how we’ll be picking up signals, the way you can pick up signals. And especially in driving revenue, and successful outcomes. One key stat, Ishan, I want to call out for everybody is, this was out by Gartner, is that by 2028, this is just in a couple of years, we know this technology is rapid-fire moving right now. 75% of RevOps tasks in our workflow management, data stewardships, and analytics will be executed by Agentic AI. So it’s putting pressure on our RevTech and Stack adoption to stay competitive, but there are some easy ways to do that. So… I think we’ll jump into some of the questions that will help everyone grasp how they can take away the material we’re talking about today, and gain some great revenue on their end.

Ishan Mukherjee:
Let’s dive.

Dina Habib OMara:
There you go. So, our first question. So, how are signals captured? Let’s start there.

Ishan Mukherjee:
Yeah, absolutely. So, agents are only good as the data it, that is fed into them, and the data effectively has 3 layers. One is… The signals come from the underlying sources. So you have public sources, so company data, contact data that you want to fall down. Second, you have private sources from business applications, so your 365, your Outlook, email, Salesforce, SAP, like, ERPs. And then you also have your internal, kind of, data warehouses. So, for us, what we’ve done is we’ve actually built, like, a connector system that connects, brings in public data, and then essentially brings in all your business applications and keeps it in the warehouse. On top of that, we create a context layer. Just bringing data in one place is not sufficient for agents. You have to make it agent-ready. To do agent-ready, you have to create what’s now called context graph. We call them kind of knowledge graph, and because that knowledge graph has a real-time understanding of your core entities, companies, people, products, activities. on top of all these sources, and you have to have a governance layer on top. So. Signals in agentic systems, like rocks, come from public. Private application and private data warehouses, where that data is made available as context that is secure and governed.

Dina Habib OMara:
Excellent. So, obviously, security, the security layers threaded throughout as well, as we talk about that. When we… when we think about this, and actually there’s another key statistic that applies to this, what you’re describing are these… this automation, the use of agentic AI to prevent a lot of the manual data entry as well. So we can pick up signals when it’s actioned, but there’s also the manual entry, where we… this can be prohibited. Like, for instance, significant investments are being made in platforms for CRM and other platforms by companies, but there is still a frustration out there of the need for manual entry, post-meeting documentation, and things like that. So… Tell us a little bit about the Agentic AI and how that can be put to use. And then we’ll get into how we analyze these signals.

Ishan Mukherjee:
Yeah, absolutely. Like, excited to kind of dive into it, because agents ultimately are here to do work. They may seem software, but they’re not software. They… it is different in software.

Dina Habib OMara:
Yeah.

Ishan Mukherjee:
Where this is focused to drive outcomes. In the journey to drive outcomes, you’re right that agents do make, kind of, legacy SaaS tools, legacy, kind of, platforms somewhat invisible, because A, they end up doing the work, so people are engaging with them directly, and then second, the data that it needs is actually not in… started in one of these kind of applications. So… To… and then, to go to your specific point, these agents do do the manual work on… of updating the systems, taking the pre-meeting briefs, doing the post-meeting follow-ups, and do that on autopilot mode. And then it frees up time so that, kind of the future-proof kind of employees are supercharged by using these agents in co-pilot mode. So the agent’s doing all this kind of maintained work in the background. on autopilot, and then we for what is kind of important, like in go-to-market, it’s about spending time with customers, and these agents are going to switch to co-pilot, and that’s where all the dramatic productivity gains are coming from. We’re seeing 20 to 40% increase in frontline productivity, and that’s coming from them freeing up time, but also making more of the time that they get back.

Dina Habib OMara:
We’re… we’re seeing that when we look at, in general, the statistics of, you know, the percent increases in, like, you know, the 50%, you mentioned 40%, like, absolute the ROIs that are involved, the actual resulting, efficiencies X times over for, you know, generating business outcome. This is great. So data estates coming together. You mentioned it’s as good as the data that is in, and one of the things that we talked about, it was talked about last year when all of this experimentation was going on, but the biggest thing was the data estate, unifying the data. So, with that coming together and securely, how… how would you recommend, what would you talk about as those steps? Got it.

Ishan Mukherjee:
Totally, right. When, driving agents across the revenue org is a… not a no-brainer, it is an existential priority, right? Like, agents worked in coding, agents worked in customer support and experience, they have to get rolled out and go to market, so… so that’s the highest priority at the board level. Then second is to drive that, every kind of commercial leader and the technology leader on the IT side, the CIO and the CRO, have three options. They can go build agents on their, kind of, legacy software platform, whether that’s CRM or what have you. They can go build their own agents, or they can bring in, kind of, premier, kind of, pre-built, kind of agent companies. And, the answer to your question is, where do you actually aggregate the data is driven by those decisions. What we have seen, and so we are technically in the third year of the agent cycle. The agent started to actually happen early 24. Right, so we’re 5 and 26, so we’re actually pretty kind of far along. In sales and go-to-market. It’s still pretty early, but in general, agents have been around for two and a half years. People have tried to bill agents on legacy systems, and they haven’t worked because the data isn’t there. So… so, what is happening, Dina, is what your CEO, Satya, has mentioned, which is… Businesses, especially at the higher end in the global 2000s, are centralizing their data in their own data fabric. That could be a warehouse, like a Snowflake, Databricks, or what have you, or it could be, like, something like an Azure data fabric. And that is where most of the data lives, and once that data’s in there, that’s where people are building their context crops and governance layer to feed their agents. Obviously, Microsoft Copilot’s a great example, and now So we are warehouse native, so we run in the customer’s environment on top of the data warehouse, and it just dramatically increases the time to outcome, because the context is there, the job is now to kind of get to work.

Dina Habib OMara:
Yeah, and that really lends itself to, I love how you mentioned how imperative… it is an imperative, unifying the entire data estate. It’s really about the democratization Of the data to give it intent. When we really… when we think about how… how important this is for… for all of us. It’s within our walls, but it goes beyond Broadly for the benefits that we can deliver to our organizations, and boosts our ROI.

Ishan Mukherjee:
Y-yeah, yeah, absolutely.

Dina Habib OMara:
Yeah, so let’s… let’s talk about… so everyone’s working with daily systems, CRM and so forth. Can you give us some examples of the CRM, like sales? product, support, and finance, like, the different type of systems where these could be used. Just to, you know, we have a variety of our audience today, so this might help. What could I do to put it to use? How could I do this?

Ishan Mukherjee:
Yeah, so my customer base is very enterprise, so essentially a billion plus in revenue, large kind of businesses, all the way up to, like, the global 2000. So, I can’t speak to, kind of, early-stage companies and mid-market, as well as probably, like, other kind of founders can. So, to be intellectually honest, let me talk about the enterprise and at-scale businesses, right? So, in an at-scale business, on an average, they might have 19 to 25, like, platforms, and then hundreds, maybe, like, close to a thousand SaaS applications. And the platforms are kind of what are household brands. You have the ERP, which could be at SAP, you could be their HR system, which is Workday. you would have a NetSuite, if you want, you got the CRM, broader in the go-to-market and the G&A, like, there’s only those many systems. On the other side, obviously, in R&D, you’ve got your GitHub or GitLab, you’ve got your dev environments. Each of these platforms now have some sort of agent framework. And generally, the agent frameworks are DIY agent builders, which is, go in that platform, how can you build thin agents, which are bespoke to a need, and a lot of them. That’s generally the landscape of building what you would call bolt-on DIY agents on large, existing kind of systems of record.

Dina Habib OMara:
Yeah.

Ishan Mukherjee:
what if… I just want to give you a timeline. On the R&D side, coding agents obviously work. They are working and going through the, kind of, the capital markets and the economy like nothing ever in humanity. So whether that’s pre-prod, prod, with SREs, QA, everything is now, kind of, agent-powered. Then it… then it moved to customer support, right? So, initially, you had pure, kind of, customer support agents, now there’s customer experience agents, like a SR or Decagon, they’re doing a phenomenal job. Now, I would say sales agents, what pre- and post-sales, that started to work about a couple of quarters. We’ve been… can obviously pioneered a lot of their underlying technology over the last couple of years, it was primarily co-pilot, but autopilot sales agents started to work, I would say, 3 months back. We actually have a huge launch coming next week with Microsoft and folks that you’ll see, like, now you can run your whole revenue systems in autopilot. I think that’s the sequence, then right after, like, is finance, accounting. So the way to kind of think about it is these… Agents are transforming existing large platforms, it isn’t necessarily creating net new platforms or GDPs, right? These are massive, existing, kind of, markets. Then, for… let’s take… double-click on one, like you said, CRM. So CRM is everything to do with, kind of, the customer relationship management, so pre-sales, post-sales, marketing, services. That market is unique in the sense that the data in CRM actually started to move out of CRM into the data warehouse, actually, five to seven years back. If you’re in a scale business, most of your dashboarding, reporting, pipeline actually doesn’t run off of CRM. It runs on Tableau or Looker dashboards running on some sort of data warehouse. for… in that domain, agents have accelerated that. Now, everybody wants to build agents, and they want to build an agent where all the data is going, which is not the CRM. Right? So, I would say every… HR data is not really leaving Workday. I don’t think ERP is data-leading SAP, so… so every space is different, but… and so in the CRM, that that’s what’s happened. And kind of the third part is, what are the options that does an enterprise have? So, for a specific function, if you want to go build agents, you can use an agent builder to go build your own agent, or just buy domain-specific agents. As an example, you can build your own coding agent, but most people will just use Cursor. You can build your customer support agent, some can, but most people eventually use Agent Force or Sierra. It’s just because as agents end up doing work, in some way, you’re hiring these agents, so the vendor and partner brings in domain expertise, they take on the liability, and also. on the deployment success, right? So… And that’s kind of where the agent space is today, which is most enterprises have an imperative to roll out agents. They have to choose between, kind of, incumbents, platforms, or build their own. And in building their own, they’re building a lot on their own, which is amazing, and then they’re kind of bringing in, kind of, domain-specific agents.

Dina Habib OMara:
So on that… on that note, as we’re bringing these in, how do your revenue agents work autonomously?

Ishan Mukherjee:
Agreed.

Dina Habib OMara:
Type that.

Ishan Mukherjee:
Yeah, that’s a great, great question. So, I would say we, pioneered, kind of, the concept of sales agents or revenue agents, but probably a couple years back. Before agents were a thing, we went out to start build, kind of autonomous revenue orgs for the Global 2000, and build warehouses and revenue agents. And what these agents are is effectively your best, AE, like your Strat AE, that can run as autonomously as an enterprise wants it to. So these agents know the lead-to-cash process. Really well. What that means in kind of agent speak is it’s a pre-built AI agent with all the tools necessary to do lead-to-cache already pre-built and trained. Then, the enterprise buyer has the… has the option to toggle between, kind of, co-pilot and autopilot. Today, if you had to go to some of our largest customers, Fortune 500 customers, they have BDRs on autopilot, renewals on autopilot, DealDesk on autopilot. But once, what actual, like, mid-funnel deal execution is co-pilot. So as the Strat and Enterprise AE is actually doing it, then all the CRM updates and forecastings autopilot. And that’s what businesses are doing, is they’re buying these agents, which are domain-specific. Again, they can go build their own agents for… if they were thin agents. If they want to bring in a domain-specific agent, they want an agent which has all the capabilities. They should be able to do everything in our case. From territory planning, to account research, to outreach, to meetings, to opportunity updates, to CPQ, everything has to be able to do. But then, as an enterprise vendor, like. we have to give control to the customer. Where do you want to switch between co-pilot and autopilot? So I think today, to going back, there’s a barbell distribution of productivity. The future-proof kind of employees in a function is usually kind of quota-carrying, AEs, AMs. They are getting dramatically more productive. On the other side of the barbell, there are functions which are getting leveraged or automated that have to be reskilled and, kind of move, move the other… and that’s a responsibility we take, which is. technology is one part, which is, hey, this is happening, how do we help, kind of, enable that? But on the other part, we take the responsibility actually retraining and reskilling folks, okay? Like, what does a modern rev ops role look like? What does a modern, kind of, BDR look like? And generally, to use a framing, it’s kind of how do you… go from being an Uber driver to an orchestrator of Waymo fleets, right? And that’s kind of where we’re headed.

Dina Habib OMara:
And that touches on a huge point, I think is becoming more and more prevalent, which is change management, the skilling, but that change management component within an organization, and shifting the behaviors is so key right now. That’s a huge component there. So how, in terms of human intervention. So, there is, you know, there is the fourth industrial revolution. We’re now into the… getting into the fifth industrial revolution, where human centricity is really a factor. We have also sustainability, and then we also have resilience as firms. But getting into the human centricity, this has been a big topic. surging now. We have AI. We… it’s here, like the internet. One of the things, though, is human oversight. So what human intervention and oversight is recommended from your standpoint? I know in panels that I’ve been speaking at, that human oversight is needed. It is always in there, like, if we’re looking at brand and voice, if we’re looking at distinction, if we’re looking at things where the human element is truly the way to set apart, there’s the human intervention. But in this revenue, in our revenue ops world here.

Ishan Mukherjee:
Yeah.

Dina Habib OMara:
What would… how would you… how… tell us about the human intervention and the importance of that in this type of application.

Ishan Mukherjee:
Yeah, absolutely, and like, I’m a pretty, kind of, hardcore engineer who became, like, I guess, an entrepreneur and ran a public company before this, and a lot of people, like, I think there’s two fallacies. One is, like, as an engineer, you can get, like, too technical and, like, obsessed about automation, and then on the business side, a lot of times, you don’t embrace, kind of. There’s the ways of technological progress. Which is unrelenting for us as humans. So, what I like is how do we evolve to be orchestrators of complex and creative systems? That’s what I kind of think about, is… especially in large organizations and businesses, the individuals who would thrive would be the ones who have curiosity, creativity, and can really leverage AI to go orchestrate and execute. Strategies that are human-led. Right, so, For example, on the sales and revenue side, if you look at a revenue leader or a good kind of rev ops or strategy leader, how can you use AI systems like Rox to actually potentially map your entire time in the world, segment them out better, stay on top of their customers, deploy that out of the field, run very specific sales plays, run deeply personal outreach campaigns, which is One-to-one at millions of, of, of… Kind of scale. And then, how do you kind of run these systems where, it is kind of high outcome, it’s creative, and you’re orchestrating this kind of fleet of agents? And you have massive, kind of massive leverage. So that’s kind of what we think about. Then, in terms of if you’re on the front line, there’s kind of fundamental work that… I think even Anthropic said, like, an enterprise and Strat relationship owner, like an AE or a CM, that would be the last thing that would get automated by AI. Like, there’s things that humans are phenomenal at, which is building deep relationships. kind of… kind of structuring deals, striking things. I think generally those… so that’s kind of our view, is the reality is, revenue and go-to-market organizations will look dramatically different. The future-proof employees would be one who have skinned the game, they carry quota, and they’re working with customers. they have to get deeper and kind of build deeper relationships, and maybe, like, handle more customers. On the flip side, there’s going to be individuals who there might not be that many BDRs or that many RevOps people, but the ones who would really thrive would be who can become these orchestrators. And that’s kind of generally, kind of, our view, but The last thing I would say is, it’s really easy to get lost in the theory of what will happen and what will not happen, and us as builders and operators, most of the things are honestly getting figured out in the arena, so, like. if you’re a builder, you gotta build with your customers. If you are an operator in a company, it’s just as I think somebody was saying before, which is, how do you actually… Move really fast, and try things, and kind of evolve, because… I strongly believe, like, the doers shape the future, so what means in the enterprise is going to be determined by the operators and the builders.

Dina Habib OMara:
Yeah, absolutely, and I think you hit the nail on the head with the human aspects, really, about the relationships. So, all right, being sensitive to time, we have about 3 more minutes. There is a quick success story I would love for you to highlight. You could either MongoDB or RAMP, you know, if you choose one to say, here was a challenge. an approach and a result, that would be fantastic. And then let’s give everyone some best takeaways that they can take back with them.

Ishan Mukherjee:
Yeah, absolutely. I think we have a kind of large suite of customers, and it’s kind of, you go to rocks.com and try it. By the way, we’ve been only in GA for a quarter and a half, and it’s been phenomenal to work with tens of, kind of, Global 2000s, so happy to share more. At a very high level, whether that’s CJ and Deepa and Mongo, phenomenal, like, executive CEO and CIO there. I think the high-level goal was. how do they make the business development function a lot more productive with AI? And there, what really thrived was these agents were able to do multi-turn, geo-specific personal conversations. to build interest from… from a top-of-the-funnel lead, and actually kind of bring them into the pipeline. And that’s kind of the main thing, and now we’re starting to work with them. Hopefully, we can consolidate a bunch of legacy SaaS tools, kind of free up a lot of, kind of productivity, and so that the XTR or the ADR teams get more productive. Once you’re in there, obviously the opportunity is to kind of manage… help them manage the whole kind of book of business, both net new, but also, like, install-based customers, so… if you’re looking for outcomes, it’s kind of increasing, kind of BDR productivity from 10% to 40%, while saving, consolidating kind of 50% of SaaS, and that’s kind of, kind of what’s been the poll. In terms of the tech, takeaways, the first thing is, like, you can’t, like, agents is a game of outcomes. If the, if top-down, from the executive and the board down, they’re isn’t really a concrete metric that you want to move. The numbers will not change. So, thinking through the North Star metric, whether that’s, in our case, revenue per rep, or pipeline, or win rates, is kind of the North Star metric. I think the second thing is, it’s a game of context. Like, unless you are one of the big labs, where you’re training your models, like, it is about, like, how do you aggregate and activate all the context that you need to make agents work? That would be the number two. Number three is deployment success. Like, driving success is not just a technology problem, a lot of it as a people problem, so you’ve got to drive that.

Dina Habib OMara:
Absolutely, and I would recap by saying always identify, like you were saying, your charter and objectives first, your business value outcomes, unite your data estate. That is number one, in terms of having that foundation, obviously with security, identity, and access management. This is not an ad hoc workflow, this is about a platform and integrating this. And again, the Align the human. the human oversight. So, thank you so much. Ishan, it was fantastic talking with you about this subject. Everyone, we hope you enjoyed and got some great takeaways, and let’s all go drive some great revenue and business. So, Julia, to you.

Ishan Mukherjee:
Yeah, thanks, Dina. Thanks, Julia.

Julia Nimchinski:
Amazing insights all around. Thank you so much, Ishan. Thank you, Dina, and what’s the best way to support you?

Dina Habib OMara:
I will say, we’re here to answer any questions, we’re on LinkedIn, and our continued education to the industry, so thank you, Ishan.

Ishan Mukherjee:
Yeah, likewise, feel free to reach out to me on LinkedIn, or email at ishan at rocks.com. If you know… if you want to see Revenue Agents work, live, just go to rocks.com and try it. We’re the only agent company that anybody can try, so we have thousands who try it daily, so go for it.

Dina Habib OMara:
Perfect.

Ishan Mukherjee:
Awesome. Right.

Julia Nimchinski:
Thanks so much again.

Dina Habib OMara:
Thank you, we’ll be seeing you soon.